Interactive storytelling and liveblog software for newsrooms · ten products, one editor
One toolkit for the whole newsroom day.
Built for what the desk actually does: breaking the news
Timepath is interactive storytelling and liveblog software for newsrooms. Ten products, one editor, one analytics layer. A liveblog for the story that just broke, a timeline for how we got here, a poll and a reader call-out inside the same article — and a dashboard that tells you where readers stopped, by name.
Skip the pitch, read the embed spec
Free account, no card. To embed on your own site on a free plan you verify your domain once — add a meta tag, we check it, embedding switches on.
This embed's JSON is CDN-cached for five seconds — s-maxage=5, with stale-if-error for 24 hours. Open your network tab.
Embed payload, liveblog cache, reader languages, brand typefaces · each with its source
Four numbers, and where each one comes from.
The three headline stats that used to sit here are gone. We could not tell you where they came from, so we took them down.
392.3 KiB
Mean static JavaScript and CSS an embed sends a reader
Down 17.3% from 474.5 KiB in four days, measured against production with a committed baseline harness, 28–31 July 2026.
5 seconds
How long the public liveblog feed is cached at the CDN
s-maxage=5, with stale-while-revalidate=10 and stale-if-error for 24 hours. Our internal QA target is that a published update is present for a reader within five seconds — a target we hold ourselves to, not a number in a contract.
11
Reader languages, keyed to the story's language, not the browser's
EN, NL, DE, FR, ES, PT, IT, NO, DA, SV, FI, in every reader-facing string across the suite.
17
Customer workspaces with their own licensed typefaces in the editor
79 licensed brand font files behind 71 @font-face rules, self-hosted on our embed origin. Counted from the editor's font picker.
Product changelog · shipped April–July 2026
If you last looked at Timepath a year ago, this is what is different.
Everything below is a real entry in a real Timepath Timeline, embedded the way you would embed one. Filter it by product — that filter is the reader-side control your audience gets, not a widget we built for this page.
Collections and GeoStories that rebuild themselves hourly from your own category page (→ Keep it current). A map story built from a liveblog, where every pin carries the quote it came from and a journalist ticks each row (→ The approval gate). Analytics rebuilt across nine products, with read-through curves, named drop-off points, and figures withheld when the data cannot carry them (→ Prove it worked). Print Studio, oEmbed on ten products, a WordPress plugin, and 17.3 percent off the mean embed payload (→ Ship it).
Liveblog software for newsrooms · Liveblog, Poll, Voices
You hit publish. About five seconds later it is on your article page.
That number is engineered, not hoped for. Updates coalesce on a 400-millisecond debounce, the cache purge is awaited rather than fired and forgotten, and the public feed carries a five-second CDN lifetime. Five seconds is the internal QA target we hold ourselves to — not a number in a contract, and we are not going to pretend otherwise.
A reader arriving at update 34 gets the whole story in one screen. The Cover Studio's key points deep-link to the exact update they summarise, or out to your own article, and AI can draft three to six of them from what you have already published — each one validated against your real published update list before it is allowed to link to anything.
Put a live poll or a reader call-out inside an update, as a tab in the card. No second embed, no second vendor, no second invoice.
At 300 updates you get a banner. At 500, one click starts Part 2 and carries the history across with the original timestamps — so a five-day story never becomes an 800-update page nobody can load.
The liveblog at the top of this page is the artefact for this section.
Interactive timeline and map storytelling · Timeline, GeoStory, Data Timeline
By lunchtime someone asks for the backstory. That is a timeline, a map or a dataset — not a 2,000-word sidebar.
Four editorial timeline formats — Standard, Relative, Map and Data — each horizontal or vertical. Eight renderers behind one editor, every one loaded only when a story uses it, so a Standard timeline never ships the mapping or charting library to your reader.
Chapter a long timeline into named groups in the Navigation Studio, so a reader jumps straight to The inquiry instead of scrolling through eight months. Readers can also filter by category, date or place.
Highlight individual words in a cover title: tap a word and cycle it through a brand-colour pill or brand-colour text. What you preview is the reader's own component — we deleted the standalone preview routes, so the embed is the only reader that exists. There is no gap between what you approve and what publishes.
Define a person once and reuse them. A minister, a scientist, a chief executive lives in your workspace library and drops into a timeline, a quiz or a map. Timepath can research a cited career chronology from the web, and is explicitly instructed to leave a milestone out rather than guess its date.
GeoStory puts the story on a map in your own basemap. A Belgian national daily's is a 25 KB recoloured style file served from our embed origin on free OpenFreeMap tiles — brand styling that costs nothing per reader, however far the story travels.
A Data Timeline imports a CSV or JSON dataset with column mapping and turns it into a chart-driven chronology.
Drop a Datawrapper chart, your own CMS video, a TikTok or a before-and-after image slider into any event: roughly 26 allowlisted embed hosts, importers for JW Player, Vimeo and api.video, and a search across your own site's images and video from inside the editor. Twelve European mastheads' video platforms are mapped by name, so a producer pastes a player URL and it resolves.
Print Studio exports the whole thing as a press-ready A4 PDF — a card grid you choose from 1×1 to 6×6, an optional cover, roughly 150 or 190 dpi. On vertical timelines you can also export it exactly as readers see it.
Polls, surveys, quizzes and reader call-outs · Poll, Survey, Quiz, Voices, Data Story
"Readers are split" is not a finding. 52 to 48 with an interval that overlaps is.
Four ways to bring the audience into a story, with the statistics attached. A head-to-head poll reports 95 percent Wilson intervals. A readers'-choice vote flags "too close to call" under five points and warns you on any category with fewer than a hundred votes — so you declare a winner when the data supports one, and not before.
Five poll formats: a standard multi-question poll with a free-text "Other", a candidate vote with an unbounded field and per-category ballots, head-to-head, drag ranking with agreement analysis, and a line-up poll on a rendered pitch.
Fifteen survey question types with Pathway branching — and the analytics separate drop-off (never reached the question) from skip (reached it and refused), a distinction most tools quietly merge into one number.
One click turns a finished survey into a published Data Story: nine block types, six starting templates, your choice of chart per question.
Voices runs a moderated public call-out. Multi-prompt, configurable on exactly what you collect, with the consent wording snapshotted onto every single submission — so a year later you can prove what a contributor agreed to. Moderation runs pending, approved, reported, rejected; readers can report; three flags auto-hide an item; and approved submissions publish straight into a Collection.
Quiz: four mechanics across three layouts is twelve reader experiences, generated from a PDF, a Word file, an article URL, an RSS feed, a category page or just a topic — and the analytics tell you which wrong answer readers actually chose.
Self-updating topic pages and dossiers · Collection, GeoStory, Social Story
The dossier page you built in March is still right in November. You did not touch it.
Point a Collection or a GeoStory at your own category, tag or topic URL. An hourly job re-reads it and pulls in what is new. You do not need to know CSS: paste the URL, describe the articles in plain language, and the selector is chosen from a census of the repeating elements on your page — then re-validated on our server before it is trusted.
Collections refresh at five past the hour. GeoStories at quarter past. New GeoStory pins land in a moderation queue with per-item and bulk approve or reject, so a map never gains a location without a person.
When your CMS changes and the rule stops matching, you get a status card that says so — success, empty, scrape error, timeout, sync error — with the last run's real outcome and a Run now button. Not a silently empty embed sitting on a live article page.
Redirected article URLs are detected and repaired instead of left as dead tiles.
The wizard shows you why zero articles matched before you save, not after.
The same feed also serves the social desk: one article becomes an on-brand carousel filling a saved template your designer already approved, with role-differentiated slides and every platform size downloadable at once. It lands as a pending draft for a person to approve.
De Volkskrant's configuration is the worked example inside our own auto-update UI.
Last refreshed HH:MM · this feed re-reads its source page every hour at five past. Come back after the hour and check.
Interactive content analytics for newsrooms · nine products, one dashboard
"Where did readers stop?" — "Event 12. Inval Arbeidsinspectie. Down 38 percent."
Nine products, one analytics layer. Every dashboard opens on the same reach numbers, then answers the question the format actually raises: a read-through curve in true reading order for a timeline or a liveblog, coverage per visitor for a map story, which wrong answer readers picked on a quiz, drop-off versus skip on a survey.
The dashboard withholds figures it cannot support. A period-over-period change only appears when the prior window sits entirely inside the tracked range. A rate above 100 percent is withheld rather than clamped. A per-visitor ratio is withheld under thirty visitors. A missing value renders as an em-dash, never as a zero.
Every section carries a provenance line: "24 of 28 events have recorded data · tracking since 21 June 2026." You can see what a number is made of before you quote it in conference.
Newsroom customer stories · RTL Nieuws and Mediahuis on covering a story live
What the desks say.
As the chief editor of RTL Nieuws, I can confidently say that with Timepath liveblogging, we're able to provide real-time updates to our readers as events unfold, keeping them informed and engaged like never before.
Dennis van LulingChief Editor, RTL NieuwsTimepath has become an essential part of our digital publishing toolkit. It allows our teams to enrich articles with interactive elements in just a few clicks that truly engage our readers. We've seen a clear boost in time spent on page and stronger reader interaction across the board. It's simple to implement, scales well, and fits seamlessly into our editorial workflows.
Han-Menno DepewegChief Technology Officer, Mediahuis
oEmbed, WordPress and white-label embedding for publishers
Paste the URL into WordPress. Or hand your CMS team three CDN requests.
Ten products publish a spec-compliant oEmbed endpoint with discovery links on every viewer page. Our WordPress plugin ships a Gutenberg block and a [timepath] shortcode — paste a Timepath URL, get a live embed, with no snippet to maintain.
Two rails. The hosted iframe, or an inline UMD script that mounts React straight into your own DOM — three CDN requests instead of seventeen to twenty-six, with the story's content living in your article's markup.
It looks like your masthead, not ours. Switch mastheads on the story beside this text: same story, three newsrooms' type, their colours, their basemap.
Human-in-the-loop AI for newsrooms · every draft waits for an editor to accept it
Our AI cannot publish. Only your desk can.
Every AI feature in Timepath produces a draft that a person has to accept. That is not a policy statement, it is how the code is built — and in one case it is enforced by the shape of the tool schema itself.
The model is not allowed to invent a coordinate.
Paste a forty-update liveblog and one instruction. A ten-stage server pipeline returns every candidate location with coordinates, country, role, date, source link, confidence and warnings. The model's tool schema physically cannot emit a latitude — only a disambiguated place name, which a gazetteer resolves. A hallucinated coordinate is impossible, not merely discouraged.
Every pin carries the verbatim quote it came from, checked as a contiguous run against the exact text the model saw. Eighty-eight of eighty-eight unit tests passing.
Partial coverage is reported, not hidden.
Failed chunks, dropped segments, truncated source and cap hits all appear in the review panel instead of being smoothed into a confident partial answer. Caps: 40 points, 400,000 pasted characters.
The same gate everywhere.
AI-drafted liveblog key points are validated against your real published update list before they can link to one. Suggested sport scores stage as a pending review with a citation and Approve or Reject, never written live. Auto-updated map pins land in a moderation queue. Generated social posts land pending.
Off switches that actually switch it off.
A kill switch per workspace and per organisation, and every call metered against your licence — free blocked, pro 30 a minute and 1,500 a day, enterprise 60 a minute and 15,000 a day.
Your desk's own review loop.
Comments live in five editors, scoped to a whole product or to a single event, open until someone resolves them, with rich text and @mentions. Timelines carry a server-written edit history: who changed what, when, per event, last fifty edits. One gap worth knowing: an @mention is stored and shown, but no email goes out yet.
A licensed, metered alternative to staff pasting copy into a consumer chatbot.
The AI workspace ships a Headline Generator, a web-grounded Real-Time Fact Checker, an Article Summarizer, an SEO Optimizer and a Tone Adjuster, scoped to your workspace with usage you can see. Not sure which format fits a story? Paste the article URL and Timepath recommends two or three — and it is structurally unable to recommend a product your workspace has not enabled. Available on request.
Timepath alternatives compared · campaign platforms, chart tools, widgets, in-house
Four things newsrooms try before this, and where each one stops.
We would rather you knew now. Three of these four are good tools. The question is what happens in month six.
A campaign or gamification platform.
- Good at
- running a competition with a prize draw and a consent flow, fast.
- Stops at
- it is built for marketing, so the output looks like a campaign rather than like your paper, and the numbers live in a dashboard your editors never open.
- Our answer
- Voices — a moderated call-out where the consent wording is snapshotted onto every submission, three reader reports auto-hide an item, and approved responses publish straight into a Collection.
A chart or map tool.
- Good at
- one excellent chart, quickly, made by a data journalist who already knows the tool.
- Stops at
- it is not a story format — no read-through, no named drop-off, no reader path — and it does nothing for the liveblog, the quiz or the poll around it.
- Our answer
- keep it. Datawrapper and LocalFocus are on our allowlist of in-story embed hosts; a chart drops straight into a timeline event.
A generic embed widget.
- Good at
- cheap, immediate, works.
- Stops at
- it turns up in someone else's typeface with someone else's logo, and it is the one element on the page a reader can tell you did not make.
- Our answer
- seventy-nine licensed brand font files, your own recoloured basemap, and your own footer line.
Building it in-house.
- Good at
- exactly what you specify, owned by you, no vendor.
- Stops at
- eight timeline renderers, twelve quiz reader experiences, fifteen survey question types, five poll formats, an oEmbed provider on ten products, a WordPress plugin, eleven reader languages and a payload budget — all of which then need maintaining while your team covers an election.
- Our answer
- build the thing only you can build. Let us maintain the format layer.
Interactive content formats for newsrooms
Ten products. One editor, one design system, one dashboard.
The count is not the point. The point is that a hard-news quiz, a readers'-choice award, a starting-XI vote and a forty-update liveblog come out of the same product, wearing the same brand kit, landing in the same analytics — instead of four vendors and four invoices. Nothing below loads until you press it, and the number on each button is the static JavaScript and CSS a reader downloads, measured against production.
Storytelling & Context
Four editorial formats, two orientations, eight renderers. Chapters, reader filters, and a printable A4 handout.
About five seconds from publish to reader. Key points that deep-link. A poll inside an update.
A topic page that rebuilds itself hourly from your own category URL.
Interact & Learn
Twelve reader experiences. Built from a PDF, a URL or a topic. Per-distractor analytics.
Five formats, from a two-option vote to a sixty-candidate readers' choice, with 95 percent intervals.
Fifteen question types with branching. Drop-off and skip counted separately.
Community & Social
A moderated public call-out. The consent wording is snapshotted onto every submission.
One article, an on-brand carousel, every platform size. Lands as a pending draft.
Visualize & Explore
Location storytelling on your own basemap, at zero cost per reader.
Nine block types, six starting templates. Import a finished survey in one click.
Measured with a committed harness against production, static JavaScript and CSS only. The ratchet that holds these numbers is a repository script, not a CI gate — nothing fails a build today. Separately, on one 205-update liveblog the HTML fell from 420.9 KB raw / 99.7 KB gzipped to 41.2 KB / 6.4 KB. That is one long liveblog, not every load.
Packages are Essential, Business, Enterprise and Full Suite. Timeline and Collections are the base in every one. Licensing is banded on Timepath pageviews — views of your Timepath content, not your site's traffic.
Procurement and vendor review · reader data, AI governance, SSO, accessibility
The questions procurement asks, answered before they ask.
Send this section to whoever reviews your vendors. Where the answer is "not yet", it says not yet.
- Reader data
Lead personal data is masked for editors, unmaskable only by a verified admin, never CSV-exportable, and anonymised in place ninety days after a product closes.
- AI governance
Every generation runs in a licensed, metered, workspace-scoped environment with a kill switch at workspace and organisation level, and nothing it drafts reaches a reader without a person accepting it. If data-processing location is a scored requirement in your tender, raise it on the first call and we will answer it in writing.
- Security findings
Externally reported findings, each tracked to a committed fix. We have not commissioned a third-party penetration test and we are not going to say we have.
- Account security
Server-side account creation, canonical email de-duplication, a vendored disposable-domain blocklist, IP rate limits on signup and verification resend, and a weighted signup-risk score. Product data is written through authenticated server routes, never straight from a browser.
- Free-tier embedding
- Read more
Free plans embed on a domain you verify with a meta tag, so a stranger cannot lift your snippet onto their own site. Paid plans skip the check.
- Single sign-on
SAML SSO is available for enterprise customers, configured by our team. It is live with RTL Nieuws.
- Accessibility
- Read more
If WCAG conformance is a scored requirement in your tender, tell us on the first call. We will answer it in writing, item by item, rather than send you a badge.
- Resilience
If we have an outage, the liveblog embed keeps serving the last good version to your article for up to twenty-four hours, and iframes fail to a skeleton rather than a broken page.
Timepath FAQ · migrating, white-labelling, desk review, page weight, data, pricing
The questions we actually get asked.
Asked in demos, answered here, same answers.
You embed a Timepath liveblog on one article and run it alongside for a week. There is no CMS migration: the embed sits in your article the way any other embed does, through oEmbed, our WordPress plugin, an iframe or a script tag.
If you decide against it, you remove the embed.
Free Timepath account · no card, embedded on your own domain
Start with one story.
Take the next thing you would cover live, or the next explainer that needs a timeline, and build it in Timepath instead. Free account, no card, embedded on your own domain after a one-time meta-tag check. If it earns its place, we will talk about a package.
